You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On current master 1c989ed
/Documentation/op-guide/supported-platform.md
says
etcd has known issues on 32-bit systems due to a bug in the Go runtime
This seems incorrect. Unless you are suggesting that all int64 in structs should be aligned to 64-bit boundaries and lead to increased memory usage for the common case. Likewise, it seems impossible for a compile to check all uses of an exported symbol to check if it is used with "sync/atomic".
Unless I'm misunderstanding the scope of the issue, an initial grep only shows 14 files that use "sync/atomic", it should be easy to do a check for alignment on these files.
At the very least, it seems incorrect to blame this on a bug in the Go runtime. I do understand there may be limits with how memory is mapped and that there isn't a builder for those archs or other limits may apply for general "support".
The text was updated successfully, but these errors were encountered:
On current master 1c989ed
/Documentation/op-guide/supported-platform.md
says
This seems incorrect. Unless you are suggesting that all int64 in structs should be aligned to 64-bit boundaries and lead to increased memory usage for the common case. Likewise, it seems impossible for a compile to check all uses of an exported symbol to check if it is used with "sync/atomic".
Unless I'm misunderstanding the scope of the issue, an initial grep only shows 14 files that use "sync/atomic", it should be easy to do a check for alignment on these files.
At the very least, it seems incorrect to blame this on a bug in the Go runtime. I do understand there may be limits with how memory is mapped and that there isn't a builder for those archs or other limits may apply for general "support".
The text was updated successfully, but these errors were encountered: